Croatian mayflies (Insecta, Ephemeroptera): species diversity and distribution patterns

Abstract Knowledge of the mayfly biodiversity in the Balkan Peninsula is still far from complete. Compared to the neighbouring countries, the mayfly fauna in Croatia is very poorly known. Situated at the crossroads of central and Mediterranean Europe and the Balkan Peninsula, Croatia is divided into two ecoregions: Dinaric western Balkan and Pannonian lowland. Mayflies were sampled between 2003 and 2013 at 171 sites, and a total of 66 species was recorded. Combined with the literature data, the Croatian mayfly fauna reached a total of 79 taxa. Of these, 29 species were recorded for the first time in Croatia while 15 species were not previously recorded in Dinaric western Balkan ecoregion. Based on the mayfly assemblage, sampling sites were first structured by ecoregion and then by habitat type. In comparison with the surrounding countries, the Croatian mayfly fauna is the most similar to the Hungarian and Bosnian fauna. Some morphologically interesting taxa such as Baetis cf. nubecularis Eaton, 1898 and Rhithrogena from the diaphana group were recorded. Ephemera cf. parnassiana Demoulin, 1958, the species previously recorded only from Greece, was also recorded.


Introduction
Mayflies (Ephemeroptera) have a worldwide distribution, being absent only from Arctic region, Antarctica and some remote oceanic islands (Barber-James et al. 2008). According to the literature (Bauernfeind and Soldán 2012), 369 species are recorded for Europe and North Africa. Mayflies are a merolimnic insect order (i.e. with aquatic larval stages and terrestrial adults) that plays a critical role in running and standing waters where they hold an important position in secondary production, as an important food source for diverse freshwater and terrestrial predators. In recent decades, human impacts on the distribution and abundance of many aquatic insects, including mayflies, are becoming more and more evident. During the 20th century, increasing industrialisation, population growth, overexploitation of natural resources and different types of pollutions have greatly impacted many European freshwater ecosystems, and also endangering the species inhabiting them (Brittain and Sartori 2009). Highly sensitive, confronted with habitat alteration, mayfly species are among the first to disappear. Therefore they are important indicators of freshwater health and widely used in bio-monitoring programmes over the world (Elliott et al. 1988, Sartori andBrittain 2015). The knowledge of the mayfly biodiversity in the Balkan Peninsula is still far from complete. Moreover, many taxa lack appropriate morphological descriptions for the larval and/or adult stages. The mayfly fauna in Croatia is no exception. Published data on Croatian mayflies are generally part of diverse limnological studies (e.g. Matoničkin 1959, 1987, Matoničkin and Pavletić 1961, 1967, Filipović 1976, Habdija and Primc 1987, Habdija et al. 1994, 2004 in which mayflies were investigated only as part of the overall macroinvertebrate fauna. In most studies, identification tools are generally not cited, thus the accuracy of mayfly species identification is questionable. In summary, 50 mayfly species were recorded from Croatia (Bauernfeind and Soldán 2012, Kovács and Murányi 2013, Ćuk et al 2015. In comparison with the number of species recorded in the neighbouring countries, i.e. 68 in Slovenia, 106 in Italy, and 93 in Hungary (Bauernfeind and Soldán 2012), it can be assumed that the Croatian mayfly fauna has been underestimated to date.
Studies on distribution and biodiversity are of crucial importance in determining the conservation status of certain species and in investigating factors that influence that diversity (de Silva and Medellín 2001). Therefore, knowledge of the mayfly faunal composition, seasonal dynamics, distribution, ecology, biogeography and especially their sensitivity as bio-indicators can enable high-quality classification and protection of Croatian freshwater habitats.

Sampling and laboratory methods
Croatia is a relatively small country situated at the crossroads of Central and Mediterranean Europe and Balkan Peninsula, and is divided into two ecoregions: Dinaric western Balkan (ER5) and Pannonian lowland (ER11) (Illies 1978). Specimens were collected in lotic and lentic freshwater habitats throughout the Croatian territory ( Fig.  1). Additionally, specimens housed in the collection of the Slovene National History Museum were identified.
The list of the 171 sampling site names with number codes (site ID), altitude, latitude and longitude is presented in Table 1 as well as on the map (Fig. 1). Larvae were sampled using a Surber sampler and hand net, adults using hand nets and pyramidal emergence traps.  Table 1 for codes).  Illies (1978); Dinaric western Balkan (5) and Pannonian lowland (11). BS = Black Sea Basin; AS = Adriatic Sea Basin. Mayflies were sampled in every season at 34 sites, while at the remainder of sites, sampling was usually performed only once between April and September. Specimens were stored in 80% ethanol and identified in the lab using a stereomicroscope and microscope. A reference collection was made by preparing permanent slide mounts of identified species. Larvae were treated with 10% KOH and 99% acetic acid to remove all muscle parts. Mouth parts, legs, gills, thorax, abdomen, paraproct plate in Baetidae and cerci, necessary for the species identification, were fixed in Euparal and examined under a microscope. Adult specimens were mostly identified by the imaginal male genitalia. The collected material (larvae and adult specimens) was identified using Müller-Liebenau (1969), Elliott and Humpesch (1983), Malzacher (1984), Elliott et al. (1988), Studemann et al. (1992), Haybach (1999), Bauernfeind and Humpesch (2001), combined with numerous publications with species descriptions (e.g. Tomka and Rasch 1993).

Data analysis
All recorded specimens were included in the Croatian mayfly species list. Data for the sites with the same sampling effort were statistically analysed using the PRIMER 6 software package (Clarke and Warwick 2001). As such, only 34 sampling sites were compared out of the total 171 (Table 1). These sites were sampled in all seasons, at the available microhabitats and they represent habitats in each ecoregion and each sea basin. Species diversity, evenness, and similarity between sites with respect to the mayfly composition and abundance were determined by the Shannon-Weaver and Simpson indices. For estimation of similarity and differences in the mayfly community composition, cluster analysis was used. Similarity among sites was determined using the Bray-Curtis similarity index. SIMPER (Similarity Percentage) was used to assess which taxa are primarily responsible for the similarities between the sites of the same habitat type. The Croatian mayfly species richness was compared with the surrounding countries (Bosnia & Herzegovina, Hungary, Slovenia, Italy) by compiling species list for these countries taken from Bauernfeind and Soldán (2012) and the Sørensen Index of Similarity was calculated.

Species richness
In total, 79 mayfly taxa (Table 2) were recorded for Croatia. Of the 171 sites (55 in ER11, 116 in ER5) investigated during this study (Table 1), 66 taxa were sampled, of which 29 were recorded for the first time ( Table 2). The presence of 13 (16%) previously recorded species could not be confirmed (  (Curtis, 1834).
Approximately half of the species (30) were present in both ecoregions. A total of 50 species was recorded as present only in the Dinaric western Balkan ecoregion (ER5) and 48 only in the Pannonian lowland ecoregion (ER11) ( Table 2). Nearly half the species (32)  (2012) -presence in Croatia noted without referent to exact localities. ▼ Only literature data: Kovács and Murányi (2013). ■ Only literature data: Ćuk et al. (2015).
were recorded in both the Black and Adriatic Sea Basins, while 25 species were recorded only for Black Sea basin and 11 species only for Adriatic Sea basin ( Table 2).
The Sørensen Index of Similarity indicated the Croatian mayfly fauna had the greatest similarity with the Hungarian assemblage (Table 3).

Mayflies (Insecta, Ephemeroptera) of Croatia
For the distribution data, the following format was used: "Literature data" were mainly taken from Bauernfeind and Soldán (2012), which listed the presence of each species in Croatia but without reference to their exact localities. Two and one species and localities where they were recorded were mentioned in Kovács and Murányi (2013) and Ćuk et al., respectively. "Literature data with new records" corresponds to data obtained as a part of this study but were already published. "New records" are data obtained in this study but were not yet published. For every species, the site ID is listed. All sampling sites and their ID numbers are listed in Table 1.

Community composition
The majority of the Croatian mayfly species were found to be associated with rivers and streams (Table 2). Among these, larvae of ten species also occurred within the spring areas (Table 2). Eleven species recorded in lakes and/or ponds were also found to inhabit flowing-water habitats. Cluster analysis (Fig. 2) showed that based on the  Table  1 for codes). mayfly assemblage, sampling sites were mainly structured first by ecoregion and then by habitat type. Species richness at the sampling sites and diversity indices are presented in Table 4. Species richness ranged from 2 and 18 species, Shannon-Weaver index between 0.21 and 1.96 and Simpson index between 0.11 and 0.82. All sampling sites with the highest species richness and diversity indices were situated in the Dinaric western Balkan ecoregion (ER5). The SIMPER analysis between sites within the same habitat type showed an average similarity ranging from 35.1% for the Pannonian lowland rivers to 57.3% for the springs (Table 5).

Discussion
Due to the paucity of systematic studies, mayfly fauna and their habitat preferences in Croatia were very poorly known, with records of only 50 species (Bauernfeind and Soldán 2012, Kovács and Murányi 2013, Ćuk et al. 2015. As expected, this study showed a higher diversity: 66 taxa were recorded, of which 29 for the first time in Croatia (Table 2). Combined with the literature, the species list consists of 79 taxa. Croatia is a relatively small Balkan country divided into two Ecoregions: Dinaric western Balkan (ER5) and Pannonian lowland (ER11) (Illies 1978) due to its position on the crossroads of Central and Mediterranean Europe, which is why its mayfly fauna shows transitive characteristics.
As a result, species with wide (e.g. Baetis rhodani, Cloeon dipterum, Caenis horaria, Serratella ignita), patchy (e.g. Procloeon nana, Leptophlebia vespertina, Caenis beskiden- Table 5. SIMPER analysis for similarities in mayfly community composition in different habitat types (Pannonian lowland river, Dinaric river, Spring, Tufa barrier, Lake). Average similarity reflects the percentage between samples within one habitat type.  (Buffagni et al. 2007, 2009, Bauernfeind and Soldán 2012. The new records include several morphologically interesting taxa: Rhithrogena from the diaphana group, Baetis cf. nubecularis and Ephemera cf. parnassiana. The Rhithrogena species from the diaphana group is morphologically similar to Rhithrogena savoiensis Alba-Tercedor & Sowa, 1987. However, DNA analysis based on mitochondrial COI gene shows it to be more closely related to Rhithrogena beskidensis Alba-Tercedor & Sowa, 1987 (Vuataz unpubl. results -Liebenau, 1969), which presents the morphological characteristics that are intermediate between Baetis alpinus and B. nubecularis. Interestingly, the species is only recorded in high numbers  in the mountain Dinaric karst streams and tufa barriers in the area of Plitvice Lakes National Park (Table 1, Fig. 1). One male imago of the genus Ephemera Linnaeus, 1758, was caught in the Lopoško vrelo stream in southern Croatia. Its morphological features correspond to Ephemera parnassiana, a species that has currently only been recorded from Greece; however due to the small sample size, additional specimens are necessary for accurate identification of the species.
As most sites were in running waters and often with a stony substrate, the most diverse genera were Baetis and Ecdyonurus, which are known to be very common in running waters of the Northern Hemisphere (Bauernfeind and Soldán 2012 ). The rare or unconfirmed presence of most of these species is likely due to the lack of seasonal sampling. It is possible that they were present at some sampling sites included in this study, but at a very young instar or even egg stage, and as such were overlooked. Additionally, some species might have become extinct from the Croatian rivers, such as Palingenia longicauda, which at present likely only inhabits the Danube River and Tisza River in Hungary, Slovakia and Ukraine (Bauernfeind and Soldán 2012).
The Black Sea basin includes 62% of Croatian rivers (Jelić et al. 2008), which likely explains the higher number of mayfly species recorded in this basin than in the Adriatic Sea basin.
The Dinaric region is considered to be a biodiversity hotspot (Bãnãrescu 2004, Griffiths et al. 2004, Ivković and Plant 2015. Despite a similar number of taxa recorded in each ecoregion, the highest species diversity was recorded for the fast flowing streams and rivers in the Dinaric western Balkan ecoregion. Similar results were obtained in the study of aquatic dance flies in Croatia (Ivković et al. 2013). The lowest number of mayfly species was found in springs and lakes (Table 4). Various studies have shown that mayfly species diversity is generally low in spring areas (Berner and Pescador 1988, Bauernfeind and Moog 2000, Maiolini et al. 2011. The only spring with four species was the spring of the Ruda River (146) in southern Croatia (Fig. 1), which is largely fed with water from the Buško Blato reservoir (Štambuk-Giljanović 2001, Bonacci andRoje-Bonacci 2003) that is relatively rich in nutrients and organic matter (Štambuk-Giljanović 2001). Thus, mayfly communities in the Ruda River spring are more species diverse and have a high proportion of detritivores (Vilenica unpubl. results). Most mayfly species prefer lotic habitats with a larger array of microhabitats, and these are less diverse in spring areas and lentic habitats. The present study confirmed the results of many previous studies (Berner and Pescador 1988, Elliott et al. 1988, Bauernfeind and Humpesch 2001, Bauernfeind and Soldán 2012. Mayfly larvae inhabit flowing and standing freshwater ecosystems where they occupy a range of microhabitats in correlation with different biotic and abiotic factors. Additionally, in running water habitats, due to the longitudinal gradient of the physico-chemical characteristics of the water, different parts of the watercourse are inhabited by different mayfly species (Elliott et al. 1988, Bauernfeind andHumpesch 2001). Cluster analysis (Fig. 2) based on mayfly assemblage generally showed that sampling sites are structured first by ecoregion and then by habitat type. For this reason, due to their morphology and water properties (Lucić et al. 2015), the large, slow Pannonian lowland rivers (Sava, Drava, Kupa) are separated from the other sampling sites situated in the Dinaric western Balkan ecoregion . SIMPER analysis (Table 5) showed that the Pannonian mayfly community consisted of species that prefer epipotamalic sections of rivers, such as Caenis macrura, Procloeon bifidum, Heptagenia sulphurea and Potamanthus luteus (Buffagni et al. 2007, 2009, Bauernfeind and Soldán 2012. Due to the two common mayfly species present in high numbers, Baetis rhodani and Rhithrogena braaschi (Vilenica et al. 2014, Vilenica unpubl. results), the investigated springs clustered together with the small mountain karst rivers. Larger karst rivers clustered together due to the presence of species with a wide ecological range as Baetis rhodani, Centroptilum luteolum, Serratella ignita and Paraleptophlebia submarginata, and species with a southern European distribution such as Rhithrogena braaschi. Another common species was Baetis liebenauae, previously recorded in smaller streams with a sandy or stony bottom as well as in large lowland rivers, where it can be found as a habitat specialist on macrophytes (Buffagni et al. 2007, 2009, Bauernfeind and Soldán 2012. The presence of a stony bottom and submerged vegetation may be a suitable habitat combination for the species. Further research is required to determine the more specific preferences at the microhabitat scale and physico-chemical properties of the water. The mayfly species diversity is generally quite poor in lentic habitats, though certain taxa can be very abundant. The main reason why lakes clustered together and apart from other sites was due to their species composition consisting of taxa from lentic (e.g. Caenis horaria) or a wide range of habitat type preferences (e.g. Centroptilum luteolum, Ephemera danica; Bauernfeind and Soldán 2012). Due to the presence and abundance of the species Baetis rhodani, B. cf. nubecularis, Centroptilum luteolum, Serratella ignita, Ephemera danica and Paraleptophlebia submarginata, the lower streams in the Plitvice Lakes National Park (sites 84 and 85) grouped together with the tufa-barriers (see also in Vilenica et al. 2014).
In comparison with the neighbouring countries and with consideration of their surface areas, the Ephemeroptera diversity in Croatia could be characterised as relatively high. Together with Croatia, Bosnia and Herzegovina is also situated in Dinaric western Balkan ecoregion (ER5) (Illies 1978). However, as its mayfly fauna is currently poorly known, with only 52 species recorded, and as a large part of Croatian territory belongs to the Pannonian lowland ecoregion, to which most of the Hungarian territory also belongs, the Croatian mayfly fauna was found to be most similar to the Hungarian fauna (75%, Table 3). This is due to the presence of widely distributed species and of the species inhabiting the larger rivers. Even though the mayfly fauna of Bosnia and Herzegovina is currently poorly known, 65% of the species were similar to the Croatian fauna. Thus, it is possible that a much greater similarity between these countries can be expected in the future. Italy is divided into two completely different ecoregions than Croatia: Italy (ER3) and Alps (ER4) (Illies 1978). It had a much higher mayfly diversity and the lowest similarity with the Croatian mayfly assemblage (55% , Table 3). This is possibly due to its geographical position and large surface area that includes a great variety of geographical features and diverse habitats. For example, the Alps, which are not present in Croatia, are well-known for their mayfly diversity and endemism, especially in the genus Rhithrogena Eaton, 1881 (Vuataz et al. 2011).

Conclusions
As expected, this study revealed a higher number of mayfly taxa inhabiting Croatian freshwater habitats than known from the previous literature. As two of the most similar mayfly assemblages of the neighbouring countries have several taxa that could also inhabit Croatian habitats (e. g. Baetis vardarensis Ikonomov, 1962, Rhithrogena picteti Sowa, 1971, Leptophlebia marginata (Linnaeus, 1767, Ephemerella notata Eaton, 1887, Caenis luctuosa (Burmeister, 1839)) but were not yet recorded, due to the lack of systematic sampling in all seasons, future studies should include seasonal sampling of a higher number of sites and habitat types. Additionally, the main focus should be on the eastern lowland part of the country, where a lower number of sites was visited during this study.
In the present study, some interesting taxa with restricted European and local distributions were recorded (e.g. Rhithrogena gr. diaphana, Baetis cf. nubecularis and Ephemera cf. parnassiana). Considering these species were recorded from a small number of sites in this study, they could be considered rare. Future studies on the taxonomic status, ecological features and detailed distribution of these species is necessary.
Additionally, as Baetis liebenauae was recorded on larger karstic rivers, a different habitat type than previously known, more detailed information on its preferences at the microhabitat scale and water physico-chemical properties should be investigated.